Genetic characteristics, clinical spectrum, and incidence of neonatal diabetes in the Emirate of AbuDhabi, United Arab Emirates

INTRODUCTION

Neonatal diabetes mellitus (NDM) is defined as hyperglycemia diagnosed in the first 6 months of life and needing treatment. NDM can be permanent (PNDM) or transient (TNDM). Using the current definition, the incidence of NDM was 1:89,000 in Germany [Grulich-Henn, 2010], 1:90,000 in Italy [Iafusco, 2012], and 1:160,949 in Austria [Wiedemann, 2010]. In these studies, the proportion of TNDM patients was 10%, 40%, and 70% respectively. The remaining epidemiological data on NDM were confined to PNDM [Bappal, 1999; Slingerland, 2009; Habeb, 2012a; Kanakatti, 2013; Abujbara, 2014; Demirbilek, 2015].

NDM has a monogenic etiology with mutations in more than 20 genes identified in over 80% of patients [Rubio-Cabezas, 2013; Flanagan, 2014]. These mutations cause diabetes through different mechanisms resulting in variable phenotypes. In PNDM the genotype and phenotype is largely influenced by paternal consanguinity. Heterozygous mutations in the K_ATP channel genes (KCNJ11 and ABCC8), are the commonest cause of PNDM in outbred populations [Vaxillaire, 2004; Suzuki, 2007; Edghill, 2008; Støy J, 2008; Rubio-Cabezas, 2013]. Most patients with these mutations have isolated diabetes and achieve better glycemic control when treated with oral sulphonylurea (SU) rather than insulin [Pearson, 2006; Rafi, 2008]. However in consanguineous families, PNDM is more likely to be syndromic with recessive EIF2AK3 mutations being the most frequent cause [Docherty, 2013; Rubio-Cabezas, 2013]. Approximately 70% of TNDM subjects have imprinting defects of chromosome 6q24 and the remaining cases are caused by mutations in KCNJ11, ABCC8, INS, or HNF1B. Patients with 6q24 TNDM typically present in the first week of life, go into remission at a mean age of 12 weeks and more than 40% have extra-pancreatic congenital anomalies [Docherty, 2013; Rubio-Cabezas, 2013]. Interestingly some 6q24 TNDM patients relapse during adolescence [Flanagan, 2014].

In the Gulf region, data on NDM are limited to two studies on PNDM: In Oman, an incidence of 1:45,000 was reported; but the authors used a cut-off age of 4 weeks to define PNDM and no patient had genetic testing [Bappal, 1999]. However a recent study, from the highly consanguineous population of northwest Saudi Arabia, reported a high PNDM incidence as high as 1:21,000 with most patients having PNDM as a part of a rare familial autosomal-recessive syndrome [Habeb, 2012b].

We aimed to describe the genotype and phenotype of patients with NDM diagnosed in Emirate of Abu Dhabi, United Arab Emirates (UAE), and estimates its incidence during 1985–2013.

MATERIALS AND METHODS

The study was conducted according to the Helsinki Declaration and a signed written consent for genetic testing was obtained from all families.

Emirate of Abu Dhabi is the largest of seven Emirates forming the UAE. Its 2.2 million populations is a mixture of Arabs and expats from neighboring Asian countries and most of them are split between the city of Abu Dhabi, which is the Capital of UAE, and Al-Ain city. The birth rate in the Emirate of AbuDhabi is among the highest in the world and consanguinity is highly practiced [Tadmouri, 2009]. The pediatric diabetes service is provided by a number of units in AbuDhabi and Al-Ain through a compulsory health insurance system regulated by Health Authority Abu Dhabi (HAAD).

RESULTS

During 1985–2015 25 patients (16 males) from 15 families with NDM were identified by 4 units. Five patients were diagnosed in Al-Ain and the rest were from Abu Dhabi. All 25 patients were pancreatic autoantibody-negative, two (8%) had TNDM and 23 (92%) were PNDM. Based on the life births of 701,807 during 1985–2013, the incidence of NDM in Emirate of AbuDhabi during this period was at least 1 in 29,241 live births, with a rate of 1:31,900 for PNDM and 1:350,903 for TNDM. A comparison of this incidence to rates in other populations is illustrated in Figure 1.

Phenotype and Clinical Management

The phenotype of the NDM patients and three subjects diagnosed after age 6 months are summarized in Table I.

All 25 patients with NDM were born small for gestational age and presented with diabetes at a mean age of 5 weeks (range from 1 hr to 25 weeks). All except three were born to consanguineous parents and 44% (11/25) had extra-pancreatic features. Of the 23 PNDM patients, 10 (43.4%) had isolated PNDM and three had pancreatic aplasia on imaging. Twins were delivered at 31 weeks while the rest were born at term. The eldest surviving patient was 30 years and 7/20 (31%) patients died of non-diabetes illnesses. All surviving PNDM patients except one are still on insulin.

Mutations in EIF2AK3 causing Wolcott-Rallison syndrome (WRS) were identified in 39.1% (9/23) of our PNDM patients. We recently described the details of these patients within a large WRS series [Habeb, 2015]. In summary, all patients had diabetes, liver disease, and short stature. Neutropenia was observed in eight children and seven needed intermittent courses of erythropoietin for normochromic normocytic anemia. Five patients had skeletal dysplasia (SD), two needed multiple surgeries and one was wheelchair-ridden. Five patients died of acute fulminant hepatitis while patient 2.1 was rescued by liver transplantation (LT) at 2.4 years. He was the only child in this cohort to present beyond the age of 6 months and 6 years post LT he had better diabetes control with no liver disease. However his SD deteriorated resulting in severe spine and lower limbs deformities. Phenotypic variation in the onset and severity of WRS features was demonstrated among six affected members of an extended family (1.1–6; Fig. 2). In this family, diabetes control was variable despite intensive insulin therapy and was complicated by recurrent unpredictable hypoglycemic convulsions which often did not respond to glucagon. Recently, the parents of patient 1.3 opted for pre-implantation Genetic Diagnosis (PGD) for WRS and delivered a healthy girl.

The nine patients with the INS c.-331C>G mutation were from two unrelated families. They had isolated diabetes with variability in the age of onset, duration of disease and insulin requirement even between siblings of the same gender. Four of the five affected members of family 6 (Fig. 2) were initially reported in 2010 [Garin, 2010]: two presented at 6 weeks while the other two developed diabetes at 13 and 18 years. Currently, all are young adults and on insulin. Another family member (a son of 6.3) was subsequently diagnosed with a plasma glucose level of 28 mmol in the first hour of life and currently needs an insulin dose of 1.5 units/kg in order to maintain areasonable glycemic control. All four affected members of family 7 (Fig. 2) presented with hyperglycemia on day 2 of life. However, patient 7.1 remitted at 6 weeks and remains off insulin at the current age of 6 years. In contrast, his 5 years old sister (7.2) still needs 1 mg/kg/day insulin MDI to maintain her HbA1c between 8–9%. One of twins (7.4) was started on insulin but died at 2 weeks because of sepsis while the other (7.3) remains on insulin at age 22 months.

Patient 8.1 presented with diabetic ketoacidosis at 6 weeks of age. She was born to consanguineous parents and found to be heterozygous for the KCNJ11 p.R201C mutation without neurological features. She was switched from insulin to oral sulphonylurea (SU) at 9 months and was off insulin after 5 weeks with a dose of Glibenclamide of 1.1 mg/kg/day 12 hourly. She had persistent diarrhea on the initial few days of treatment with SU requiring hospital admission. On an increased SU dose she developed postprandial hypoglycemia which improved gradually. Eight weeks after withdrawal of insulin her HbA1c was 6.4% compared to 10–12% while on insulin. Currently she is on Glibenclamide 0.4 mg/dg/day in two divided doses, has normal growth and age appropriate developmental milestones. Patient 9.1 had isolated PNDM and was heterozygous for a novel ABCC8 variant, p.H410Y. He has normal growth and intelligence and has not been tried on SU. At the age of 17 years, his insulin requirement is over 1 unit/kg/day on insulin pump therapy. Interestingly his 45-year-old mother, who is a carrier for this variant, has glucose intolerance with HbA1c of 5.9% on diet control.

The child with 6q24 TNDM (10.1) presented with hyperglycemia at 6 weeks of age and remitted 3 weeks later. He relapsed at 11 weeks requiring intermittent insulin for 18 months. He is currently 8 years old and has been off insulin. He has global developmental delay, brain atrophy and dysmorphic features (Table I). In addition he has nystagmus, agenesis of corpus callosum, and absent septum pellucidum.

Patients 11.1 and 11.2 are first degree cousins. Both presented with neonatal diabetes and features of pancreatic agenesis on MRI. Deletion of the PTF1A enhancer was detected in the younger cousin. All four children without identifiable genetic cause in the cohort had PNDM. Patient 12.1 had early features of exocrine pancreatic deficiency with absent pancreas confirmed on CT scan. He was managed by dietary support, pancreatic enzyme supplement, and intensive diabetes treatment via insulin pump. Subject 15.1 had Down syndrome confirmed by chromosomal analysis and developed non-autoimmune diabetes at 4 months. The remaining two patients (13.1 and 14.1) had isolated PNDM with good control on insulin.

DISCUSSION

We report the first data on NDM from the Emirate of AbuDhabi in the UAE. In this cohort, 84% of patients had identifiable genetic etiology with recessive EIF2AK3 mutations being the most frequent cause, followed by homozygous INS mutations. 92% of the 25 patients had PNDM and 44% showed extra-pancreatic features. The incidence of NDM in Emirates of Abu Dhabi during 1985–2013 was 1 in 29,241, which is three times the highest reported rate in Europe [Grulich-Henn, 2010]. Our PNDM incidence of 1:31,900 is comparable to the incidence of 1:48,000 reported in South East Turkey [Demirbilek, 2015] and 1 in 21,000 from Northwest Saudi Arabia [Habeb, 2012a]. However it is more than six times the incidence of PNDM in Jordan [Abujbara, 2014], the USA [Kanakatti, 2013], and three European countries [Slingerland, 2009]. We suspect that the main reason for the high incidence of this monogenic condition in our area is the combination of high consanguinity and birth rate. Of note, all except three patients were born to related parents and three of our 15 families had at least four affected members.

Earlier reports on NDM from the Gulf region focused mainly on PNDM [Bappal, 1999, Habeb, 2012a,2012b]. In our series two children had TNDM making a total number of six TNDM patients from the Gulf region reported with TNDM to date [Dib, 1995; Attia, 1998]. It is possible that in this region, TNDM is less frequent, under diagnosed or under-reported. One of our two patients with TNDM had methylation defect of 6q24 and showed most congenital anomalies described with this neonatal diabetes subtype [Diatloff-Zito, 2007; Docherty, 2013]. However, to our knowledge, his midline brain defect has not been previously reported in 6q24 TNDM. The other TNDM patient was the only of our nine patients with homozygous INS c-331C>G mutation to have TNDM. This mutation is located in the promoter region of INS and was previously found to cause TNDM in one subject in the series published by Garin, [2010]. All patients with this mutation had low birth weight in keeping with reduced insulin secretion in utero; however there was a variability in the age of onset, severity and duration of diabetes even between siblings of the same gender. Strikingly, in family 6, a boy developed hyperglycemia in the first hour of life while his father presented at 18 years and in family 7, a boy had TNDM while his 5 year old sister was still on insulin. The exact cause for this variation is unclear. Various experimental/epidemiological studies demonstrate profound influence of maternal stressors and intrauterine environment on gene expression in the placenta [Ciprian, 2010]. We speculate that factors such as epigenetic variation, intrauterine stress, or the presence of unidentified modifier genes may contribute to the phenomenon of the variability of gene expression.

WRS was the commonest cause of PNDM in our cohort as reported in other consanguineous populations [Habeb, 2012a]. This syndrome is caused by recessive EIF2AK3 mutations and characterized by early onset diabetes, intermittent hepatitis, and skeletal dysplasia. Consistent with other reports [Ozbek, 2010; Al-Shawi, 2013; Habeb, 2013], there was variability in the phenotype in the 10 patients. Most patients with WRS die during childhood with AFH and the mortality rate in our cohort was 60%, which is similar to rates reported from KSA and Turkey [Ozbek, 2010; Habeb, 2013]. The encouraging results of our first patient to undergo LT [Habeb, 2015] and the recent report of successful liver, pancreas, and kidney transplant in another child with WRS [Tzakis, 2015] raises the hope for a better life expectancy for these children. Similar to other inherited conditions with poor prognosis, genetic counseling is a cornerstone in the management of WRS. Advances in reproductive medicine enable parents at high risk of monogenic conditions to have unaffected children by selecting healthy embryos using PGD [Brezina, 2015]. This technique is less invasive than antenatal genetic diagnosis via chorionic villous sampling and avoids the need for termination of pregnancy, which is not easily acceptable in certain cultures. It provides an opportunity for carriers of recessive conditions to have a healthy child. We report the first parents of a WRS patient to have a healthy girl following PGD. The positive experience of this family would open the door for other WRS families to have healthy children.

K_ATP mutations (KCNJ11 and ABCC8) are the commonest cause of PNDM in outbred populations; however they were not found to be a common cause of PNDM in Arabs [Habeb, 2012b] or consanguineous probands with PNDM tested in Exeter [Rubio-Cabezas, 2013]. In our PNDM series 2/23 (8%) had K_ATP mutations. Given the frequency of these mutations in Europe this number is expected for our cohort of 701,000 live births. In our patient, the heterozygous KCNJ11 mutation was not detected in her related parents indicating that the mutation is either de novo or one of the parents may have germline mosaicism. Prior to this girl only one child with a KCNJ11 mutation from the Gulf region was reported to be successfully switched from SU to insulin [Al-Mahdi, 2010]. However that child was born to an unrelated couple. The fact that our patient was born to first cousins highlights the importance of testing for KCNJ11 mutations even in NDM patients born to consanguineous parents and in populations where K_ATP mutations are not a common cause of NDM. It is not clear whether the ABCC8 p.H410Y variant identified in the patient 9.1 is pathogenic; a functional study is needed to clarify this point. A trial of SU following the functional study results will be considered.

We were able to define the genetic cause of diabetes in more than 80% of our NDM patients which is similar to rates reported in other studies of highly consanguineous populations [Habeb, 2012a; Demirbilek, 2015]. Although some patients were diagnosed with diabetes long time ago, the genetic diagnosis of NDM was confirmed recently perhaps reflecting easier access to genetic testing and better awareness by clinicians of the importance of genetic diagnosis in the management of NDM. The clinical utilities of genetic testing in our cohort were illustrated in a number of families; first it enabled us to make the diagnosis of neonatal diabetes due to a heterozygous KCNJ11 mutation and the child subsequently achieved better glycemic control following the transfer from insulin to oral SU. Second, the detection of a recessive EIF2AK3 mutation in patient 1.4 with WRS enabled his heterozygous parents to have a healthy child through the PGD. Finally, the identification of homozygous INS mutation in the proband of family 6 revised the diagnosis of diabetes in his siblings who had been diagnosed with T1DM and led to early detection of the cause of hyperglycemia in one of the offspring in the first hour of life.

Our study has two limitations: first it does not reflect the whole country but given the high birth rate and consanguinity in other parts of the UAE, a similar picture is expected. Second, patients were identified retrospectively from follow-up records of diabetes clinics; it is therefore possible that some patients were missed either due to moving to different areas or dying before the diagnosis of NDM was established. Clearly, establishing a national diabetes registry will provide more accurate figures on NDM in the whole country. In summary, in AbuDhabi, TNDM is rare. However, the frequency of PNDM is among the highest in the world and its etiology is different from Europe and the USA, but similar to Northwest Saudi Arabia and Turkey. The results of genetic testing had significant implication on clinical management of our NDM patients.